Games with no earned runs allowed by either team
Posted by Andy on February 3, 2011
Going back to the 2001 season, there have been 44 games featuring no earned runs allowed by either team. As you might imagine, these games typically feature an error allowing one or more runs to score.
Since these games tend to be very low-scoring, I would think the likelihood of going to extra-innings would be somewhat higher than usual. Of those 44 games, just 7 (16%) went to extra innings. Compare that to all 2010 games, where 220 out of 2430 games (9%) went to extra innings.
February 3rd, 2011 at 9:04 am
I know that one of them was Halladay's perfect game from last season against the Marlins.
February 3rd, 2011 at 10:08 am
I got the 2001 Yankees/Red Sox game right, as I knew that had to be on the list (that was the game where Mussina was one batter away from a perfect game).
February 3rd, 2011 at 10:27 am
Wait a minute.... it just hit me that this exposes another huge flaw in the Win-Loss crediting system — in a game where none of the pitchers allow any earned runs, how can it be rationalized to credit any pitcher with a loss? Shouldn't the fielder who made the error that allowed the winning run to score, get the big L?
February 3rd, 2011 at 11:00 am
Devon, one could make the somewhat-reasonable argument that errors are part of the game and that a pitcher who fails to pitch around it deserves the loss in that case, as compared to a pitcher who overcomes an error by his defense. Of course, this is an argument within the scope of a stupid overall system of assigning wins and losses, so it's practically not worth discussing.
February 3rd, 2011 at 12:09 pm
Devon & Andy:
I suspect it is quite random: for most pitchers, it balances out and they win as many as they lose due to unearned runs. However, as with anything involving randomness, some pitchers will have been lucky and some will be unlucky. I'm sure the distribution would be quite normal, with even the outliers quite explainable, i.e., pitching for a bad defensive team. Likewise, outliers on the "lucky" side might be pitchers for teams that used speed to disrupt defenses and force errors.
February 3rd, 2011 at 12:09 pm
Geesh. The Marlins played in 5 of the 44 games. The Yankees? Once.
As for Devon's point, I think we can all agree that W-L is not a good statistical measure of a pitcher's performance.
February 3rd, 2011 at 12:29 pm
To Howard's response we have to agree, the AL Cy Young winner was barely a .500 pitcher. If we were only looking at win-loss %, CC Sabathia would have been the Cy Young winner. Makes you wonder with all the NEW stats we recognize, how many pitchers like Nolan Ryan or Bert Blyleven might have won a Cy Young, if we were taking into account that they pitched well, for some very bad teams.
February 3rd, 2011 at 1:18 pm
I think Sandy Koufax's perfect game fits into this category.
February 3rd, 2011 at 1:29 pm
@7, Nolan Ryan's 1987 season is a great example of this. Led the league in ERA (2.76), ERA+ (142), K (270), H/9 (6.5), K/9 (11.5) and K/BB (3.10) while pitching 211.2 innings. He finished 5th in WAR for pitchers but pitched the fewest innings of the top 5 (he was 40 years old). However,
WAR for pitchers/100 innings:
2.53: Bulldog Hershiser
2.62: Bob Welch
2.30: Mike Scott
2.36 Rick Sutcliffe
2.60 Nolan Ryan
And he went 8-16 and took 5th in the Cy Young voting.
February 3rd, 2011 at 1:39 pm
So this may be a bit off topic, but it's something I've been curious of for awhile. We all know that the win/loss system is flawed. It's been beat to death sooooo many times. My question is this:
What is a better alternative? Does anyone have an alternate way to assign wins and losses to pitchers that would be better than the current system? Should we just get rid of the stat (for individual pitchers, obviously not for teams) altogether?
I'd like to hear your thoughts on this B-R community.
February 3rd, 2011 at 1:55 pm
The problem with wins is it is too dependent on your teams. There are some games where both pitchers deserve a win or both pitchers deserve a loss. You would need to basically define some level of performance that a pitcher needs to meet and then give them a win, loss, or no decision based on their stat line regardless of the score. You would also need to account for defense.
February 3rd, 2011 at 2:12 pm
@7, "how many pitchers ... might have won a Cy Young, if we were taking into account that they pitched well, for some very bad teams."
Here are a few recent pitchers who led their league in pitcher WAR while toiling for mediocre-to-bad teams, and did not win the Cy Young Award:
-- Kevin Brown, 1996 Marlins (80-82), 17-11, NL-best 1.89 ERA.
-- Jose Rijo, 1993 Reds (73-89), 14-9, 2.57.
-- Roger Clemens, 1992 Red Sox (73-89), 18-11, AL-best 2.41 ERA.
-- Ed Whitson, 1990 Padres (75-87), 14-9, 2.60 ERA (3rd).
-- Orel Hershiser, 1989 Dodgers (77-83, last in scoring), 15-15, 2.31 ERA (2nd), 257 IP (1st).
-- Orel Hershiser, 1987 Dodgers (73-89, last in scoring), 16-16, 3.06 ERA (3rd), 265 IP (1st).
-- Teddy Higuera, 1986 Brewers (77-84, 12th in scoring), 20-11, 2.79 ERA (2nd).
-- Dave Stieb, 1982 Blue Jays (78-84, 12th in scoring), 17-14, 3.25 ERA (5th), 288 IP (1st).
P.S. Stieb led the AL in pitcher WAR in 1982, '83 and '84; in '85, he was 2nd in WAR and won the ERA crown. He averaged 275 IP in those 4 years, with a 2.91 ERA and 148 ERA+. His ranking in the Cy Young votes? 4th, nada, 7th and 7th.
February 3rd, 2011 at 2:56 pm
@10, Scoop -- I heartily second your motion for a discussion on replacing "Wins" and "Losses" as the first-order counting stats for starting pitchers.
I would suggest Quality Starts (with a new definition), High-Quality Starts, and Failed Starts.
My definitions:
Quality Start:
(a) 5.0 IP to 5.2 IP, no more than 1 run allowed;
(b) 6.0 IP to 6.2 IP, no more than 2 runs allowed;
(c) 7.0 IP to 8.2 IP, no more than 3 runs allowed; and
(d) 9.0+ IP, no more than 4 runs allowed.
High-Quality Start:
(a) 6.0 IP, no runs allowed;
(b) 6.1 IP to 6.2 IP, no more than 1 run allowed;
(c) 7.0 IP to 8.2 IP, no more than 2 runs allowed; and
(d) 9.0+ IP, no more than 3 runs allowed.
Failed Start: Less than 5 IP.
Three points I'd like to stress about these definitions:
-- The current definition of Quality Start (6+ IP, no more than 3 ER) has the virtue of simplicity, but that is also its fatal flaw in gaining wide acceptance. Reactionaries will always criticize it based on the minimum qualification: "Six innings and three runs? That's not my idea of quality!" Using different levels of IP and runs allowed takes away that simplistic criticism; I think most people would agree that any of my 4 levels of Quality Start actually reflects a solid job by the starting pitcher.
-- These definitions are based on runs allowed, not earned runs. I think the distinction is almost moot due to the historicaly low number of errors charged in today's game and the inconsistency among official scorers. And besides, most of the help or harm from the fielders occurs in the form of range factor; so if our pitcher counting stats don't measure that, why bother measuring the effects of errors?
-- These definitions put a premium on complete games and longer outings by SPs, for two reasons: (a) I believe they're truly valuable, and (b) if managers are going to let the stat drive the strategy (as has happened with Saves), maybe this can help drive it back in the opposite direction.
(I'll try to come back later with some sample 2010 numbers based on these definitions.)
February 3rd, 2011 at 3:29 pm
2010 "records" based on my proposed definitions @13 above:
(Q = Quality Start; HQ = High-Quality Start; F = Failed Start
Felix Hernandez: 25 Q, 19 HQ, 2 F
(Actual W-L record: 13-12)
C.C. Sabathia: 21 Q, 15 HQ, 1 F
(Actual W-L record: 21-7)
February 3rd, 2011 at 3:40 pm
So 5 IP 6 earned runs is not a fail?
February 3rd, 2011 at 4:26 pm
@13 Interesting post with a good look into the quality start. I agree that the current definition of a QS is flawed and that, if corrected, it could become a better gage of pitching performance. However, you didn't really address the main part of my question. Perhaps my question was a little convoluted, so let me rephrase it:
Should the practice of assigning wins/losses to an individual continue? Or should wins and losses only be a team stat?
If you think this practice SHOULD continue, what is a preferable alternative to the current system?
We all know that the current win/loss stat is flawed, but I don't think I've ever heard someone pitch an alternate system, so I'm curious to see what we get...
February 3rd, 2011 at 4:27 pm
@15, Topper009 -- Obviously, 5 IP, 6 runs (I don't deal with ER) is not a good performance. Should it be a "Failed Start"? I don't know. I didn't want to deal with all the varieties of bad starts. If I say that 5 IP, 6 R is a "Fail," then what do I say about 0 IP, 6 R? -- or 5 IP, 12 R? And what about 7 IP, 7 R?
I'd rather leave those things to be reflected in Run Average. And so I think I'll change the term from "Failed Start" to "Short Start," to reflect the fact that it is measuring IP only
February 3rd, 2011 at 4:59 pm
@17. John, I sense you are thinking that if manager chooses to leave a pitcher in the game longer than 5 innings and the pitcher gets bombed (or has already been bombed), then it's really the manager who should shoulder the bulk of the associated blame (if there is any). Thus, if a pitcher delivers 5 innings of work, he must have been judged to be pitching well enough to last that long, and therefore his outing was not a failure.
Seems a reasonable approach.
Probably, some sort of criteria could also be developed for comparable QRA, HQRA and FRA metrics (RA=relief appearance). I've always been bugged that a relief guy can get a win or save throwing one or two pitches, or (for example) take a loss or a BS after inducing a weak grounder in hgh pressure situation that his fielder then boots.
February 3rd, 2011 at 5:12 pm
Since 1920, there have been 4 games in which both teams allowed 3 or more runs but neither team allowed an earned run.
February 3rd, 2011 at 5:16 pm
Scoop @16 -- Sorry. I should have said two things explicitly:
1. No, we should not continue assigning wins/losses to individuals.
2. I suggest Quality Starts and High-Quality Starts as the main counting stats for starting pitchers.
I can't see any way to salvage the concept of individual "Wins" and "Losses" for pitchers, while also retaining the characteristic that every game must have a "winner" and a "loser."
February 3rd, 2011 at 5:31 pm
John Autin - I like the improved Quality Start idea. I don't know if Wins and Losses will ever go away, as they are so ingrained in baseball history BUT I've always thought Quality Starts should be featured more prominently for starters and I like the different levels of QS's you propose.
Looking at your system and Quality Starts, I randomly decided to look at CC and Felix as they were mentioned earlier, across their entire careers:
**Felix: 98 of 172 career starts have been QS's (57.0%)
11 of 172 have been Short Starts (6.4%)
**CC: 182 of 322 career starts have been QS's (56.5%)
28 of 322 have been Short Starts (8.7%)
Wish I had time for more contemporaries but alas, I gotta run...
February 3rd, 2011 at 6:13 pm
Just want to add...
One postseason game where this occurred was the classic Game #5 of the 1996 World Series, won by the Yankees behind 8 1/3 shutout innings from the soon-to-be-retired Andy Pettitte.
February 3rd, 2011 at 6:15 pm
sorry, I'm stupid, your list is since 2001. if I knew how to retract my comment, I certainly would!
February 3rd, 2011 at 6:19 pm
Another index of the greatness of the 1922 St. Louis Browns: They played 13 games that season in which they allowed three or more runs but zero earned runs, and went 8-5 in those games. No other team since 1920 has played more than six such games in a single season.
February 3rd, 2011 at 6:27 pm
John Autin, Dave V. and others: why not simply use Game Score to determine the quality of a start?
February 3rd, 2011 at 6:38 pm
@13 and "These definitions are based on runs allowed, not earned runs. I think the distinction is almost moot due to the historicaly low number of errors charged in today's game and the inconsistency among official scorers.".
I agree on using runs, instead of earned runs. But, I believe the difference between the two is more than you might think. In 2010, for example, unearned runs per game were 0.37 in NL and 0.32 in AL, 8.5% and 7.2% of runs scored, respectively. This seems quite large given the overall rate of errors on fielding chances.
I suspect part of the reason for the relatively large number of "unearned" runs is the practice of regarding as uneraned all runs scoring in an inning after 3 outs would have been made (in theory), if not for the error or errors. An extreme example of this is a Mets game I remember from maybe 25 years ago where the Mets lost 12-5 or something, but did not allow a single earned run in the game.
A more reasonable definition for unearned runs would be:
- those runs currently designated as unearned; AND
- which scored on the play where the error was made
Thus, if an error is made but a run doesn't score on the play, then the pitcher and the defense have to suck it up, and play around the error. If runs score later because the error prolonged the inning, then those should be earned runs. Life isn't always fair - sometimes you have to overcome adversity. Should be the same in baseball.
February 3rd, 2011 at 6:40 pm
Andy prefaced this post by saying:
These games typically feature an error allowing 1 or more runs to score.
I wonder if these 44 games any any since 1920 had NO errors but all runs in the games were "Un-Earned".
Remember any runs that score as the result of a Passed Ball are considered UN-Earned against the pitcher.
A passed ball is not considered an error. But counts as un-earned if runs result.
One of the most famous Passed Balls was committed by Mickey Owen of the Dodgers in the 1941 World Series Game 4. Between the New York Yankees and the Brooklyn Dodgers.
Dodgers down 2 games to 1 to the Yanks led 4-3 at Ebbits Field in the Top of the 9th. Hugh Casey on the mound for the Dodgers.
Casey got Johnny Strum and Red Rolfe on quick ground outs.
Next batter was Tommy Heinrich,
On an 0-2 pitch Heinrich struck out but catcher Mickey Owen let the third strike get away, on what should have been the final out of the game
Heinrich was safe at first.
Joe DiMaggio singled, Charley Keller doubled, Bill Dickey Walked, Joe Gordon doubled, Phil Rizzuto walked, and finally Johnny Murphy grounded out.
But the Yanks had scored 4 runs in the top of the 9th. after the game was seemingly won.
P.S.
I note on the BB-Ref Recap of the game they have an E-2 listed on the line for Heinrich.
Possibly at the time a passed ball was considered an error.
Naturally now I'm starting to doubt myself of the Passed Ball not being an error???
I'll have to look up the scoring Rule on that...
February 3rd, 2011 at 6:41 pm
For CC and King Felix in 2010:
GS CCS FH
90+ 0 1
80-89 3 5
70-79 5 11
60-69 8 3
50-59 5 7
40-49 8 4
30-39 3 1
50 (QS) 21 27
>60 (HQS) 16 20
>70 (VHQS) 8 17
Sorry about the formatting!
February 3rd, 2011 at 7:08 pm
@27. Mike, exactly what I mean in 26 about not counting all runs after a 2-out error as unearned. Leaving aside the passed ball point for the moment, assume exactly the same sequence of events as you described except Tommy Heinrich reached on a booted grounder. Then, the 4 runs that scored would all be counted as unearned. Seems to me the offense wasn't given those 4 runs - they "earned" them with their clutch, 9th-inning, 2-out batting.
February 3rd, 2011 at 7:09 pm
@28. No contest. Felix is the man.
February 3rd, 2011 at 7:13 pm
Per my post {above} at @ 27:
Wikipedia backs me up that a Passed Ball is not an error
http://en.wikipedia.org/wiki/Passed_ball
Then it gives a link to the Baseball Rule Book at Section 10.13.
Reading that rule and the one above it 10.12 on assigning errors, the rules are a little vague so again I'm not sure.
However all checks via Google have assured me that a Passed Ball in not an Error, just a separate Stat Category for Catchers.
Why they gave Mickey Owen an error in the 1941 world series on a passed ball?
Well checking the BB-Ref.com Box Score again::
It looks like they never credited Owen with a Passed Ball.
Maybe, at the time they were considered Errors and no Passed Balls were assigned.
If the rule was changed down through the years or since 1941 I can't find that out.
February 3rd, 2011 at 8:57 pm
What if we were to use WPA and assigned the win to the pitcher most deserving of the win or loss- or perhaps even better, the PLAYER most deserving, regardless of position. Then, we could have an indicator of who wins/loses the most games. Of course, for this to really work, we would need to better know how to assign WPA to defensive events, so that defensive ability was factored in, but even without such analysis, it seems to me that WPA would be a good place to turn.
February 4th, 2011 at 12:02 am
@25, Bluejaysstatsgeek Says: "why not simply use Game Score to determine the quality of a start?"
BJSG -- I don't like Game Score for this purpose, mainly because of the reward for Ks and penalties for hits & BB. Don't get me wrong -- I consider Game a reasonable measure of dominance (for lack of a better term). But ultimately, the pitcher's objective is to prevent runs, not strike out opposing batters and keep them off the bases. And since I'm proposing a counting stat to replace Wins & Losses, I want to keep the same agnostic attitude towards baserunners and strikeouts that Wins & Losses have.
Consider two pitchers:
-- Alfa goes 7 innings, allowing 0 runs on 6 hits, with 0 Ks and 2 walks. Game Score: a modest 64.
-- Bravo also goes 7 innings, allowing 3 runs (2 earned) on 1 hit, with 13 Ks and 4 walks. Game Score: 74.
I personally don't accept that Bravo's game is better from a "baseball outcome" angle -- but more importantly, I don't think anyone outside the sabermetric world would buy that notion. And while these example games may not be everyday occurrences, they are far from the most extreme examples one could produce.
And putting everything else aside, Game Score is too "mathy" for the average baseball person -- not that they couldn't calculate a Game Score, but I feel certain that they wouldn't want to do it. Game Score, as you know, has 8 separate factors: a base of 50, plus 1 per out, plus 2 per complete IP starting with the 5th, plus 1 per K, minus 2 per hit, minus 4 per ER, minus 2 per UER, minus 1 per BB.
I see no chance that Game Score would be accepted as a replacement for Wins & Losses, without a major overhaul of its weights.
February 4th, 2011 at 12:31 am
Re: the discussion of the Mickey Owen WS game:
(a) MLB Rules 10.12 and 10.13 make it clear that a passed ball on a 3rd strike is not an error.
(b) Rule 10.16, "Earned Runs and Runs Allowed," says, in effect, that all runs are unearned after the 3rd out should have been recorded. And by the logic of this rule, the runs that scored after Owen's passed ball on a 3rd strike prolonged the inning, should have been unearned.
However, I've read and re-read Rule 10.16, including all the comments and examples, and I can't find a specific mention of passed balls in the context of prolonging the inning (a phrase that is not actually used in the rule, BTW) and thus leading to unearned runs. Here are the only two sentences I can find in 10.16 that touch on passed balls:
"In determining earned runs, the official scorer shall reconstruct the inning without the errors (which exclude catcher's interference) and passed balls, giving the benefit of the doubt always to the pitcher in determining which bases would have been reached by runners had there been errorless play."
"No run shall be earned when the scoring runner's advance has been aided by an error, a passed ball or defensive interference or obstruction, if in the official scorer’s judgment the run would not have scored without the aid of such misplay.
(emphasis added in both instances)
Examples are given of runs being unearned when the third out should have been recorded already, but was not because of error. No such examples are given in regard to passed balls.
Maybe I'm missing something. But it wouldn't be the first time that the rules fail to address an unexceptional situation.
http://mlb.mlb.com/mlb/official_info/official_rules/official_scorer_10.jsp
February 4th, 2011 at 1:17 am
@32, Dr. Doom re: using WPA to assign Wins & Losses:
I have grave reservations about WPA in general, and even more so about using it to assign Wins & Losses. Because it's late, I'll sum up my problem with WPA as follows:
I think it overvalues what happens in the late innings, because it simply measures the net change in the probability of one's team winning from game situation A to game situation B. And the WPA assigned to batter events is indifferent to the batter's actual merit in the event.
Consider two batters on the Mammoths. The team fell behind by 4 runs in the 1st inning, but chipped away and won it in the 9th:
-- Abel goes 3 for 4 with 3 RBI. His hits come in his first 3 times up, each one driving in a run that narrows the gap but still leaves his team behind. Each of these events has a modest positive WPA, because his team still trails; their chance of winning is better each time, but still poor overall. In his 4th trip, he strikes out in the 9th, with the game tied, 1 out and the winning run on 3rd; the WPA for this event is a large negative.
-- Baker goes 0 for 4 with no RBI. His first 3 trips are uneventful outs, each with 2 out and the bases empty; each of these events has a small negative WPA because it's a low-leverage situation. In his 4th trip, in the 9th with 2 out and a man on third, he reaches on a dropped third strike as the winning run scores; this event has a huge positive WPA. (WPA doesn't care how he reached base; it only looks at the situation before and after he batted.)
Baker ends up with a higher WPA for that game. But I think Abel contributed far more to his team's win.
(P.S. In game 4 of the 1941 WS, Tommy Henrich's WPA for striking out with 2 out in the 9th and his team down by a run was a positive 0.04, because the ball got away from Mickey Owen and Henrich reached first. That's the same positive WPA that Kirby Higbe got for a leadoff single in the 3rd with his team down a run. This strikes me as absurd on its face.)
February 4th, 2011 at 1:51 am
One more example of what I consider to be dubious valuation of batter events by WPA:
2001 WS, game 7: Through 6 innings, the Yankees didn't sniff a threat against Schilling (he faced the minimum 18 batters), but they scratched out a run in the 7th to tie the game at 1. Alfonso Soriano leads off the top of the 8th with a HR to put the Yanks in front; the WPA value is 0.23. (WPA doesn't care that Mariano Rivera, virtually invincible in the postseason, will enter in the bottom of the 8th; nor that Randy Johnson will come in later that inning.) New York doesn't threaten again the rest of the game.
In the bottom of the 9th, 2 out and men on 1st & 2nd, Tony Womack doubles down the RF line, plating the tying run and sending the potential winning run to 3rd; the WPA value is 0.50.
You can call me crazy, but I think that Soriano's HR was more valuable than Womack's double. First, Soriano created that run all by himself, while Womack got extra WPA credit by virtue of 2 batters reaching base ahead of him. Second, Soriano's HR actually put his team ahead, with 6 outs to go and Mariano coming in carrying a streak of umpteen consecutive postseason saves; granted, Womack's RBI came with the D'backs down to their last 2 outs, but it still only tied the game.
I'm not saying that WPA isn't a reasonably accurate measure of a team's odds of winning the game at a given point. If the math says that Womack's hit created about twice as big a change in the odds as Soriano's, I accept that. But I just can't accept equating WPA with player merit.
February 4th, 2011 at 1:53 am
Self-correction to my #36: I initially said that Womack batted with 2 out in the 9th; actually it was 1 out, as I noted later.
February 4th, 2011 at 2:22 am
@34 The way I read that is that errors and passed balls are treated the same by the scorer in determining earned vs. unearned runs.
Back to what the topic diverged to--should we abolish wins and losses for pitchers?
Not at all. Wins and losses are useful stats. Using them as an indicator of more than they are saying has been a problem, but that's true of any stat. People act like the stat was always very misleading, but even a couple generations ago pitchers completed a lot of games and thus their won-loss generally correlated pretty well with (not how well they pitched, necessarily) whether they allowed less runs than their opposing complement.
We have no idea how baseball will change over the coming generations, but I don't think it unreasonable that IP/start and CG may trend upwards again, strengthening the relationship between won-loss record and and responsibility for the outcome.
Also, it'd be nice to have something like the expansion of the quality start idea into narrow subsets as was mentioned above, but I'd actually do it a bit differently:
Firstly, no QS for anything below 6 innings. I don't want a five-inning pitcher.
Secondly, QS, HQS, and FS are fine for a simple system. It was mentioned that it does not address poor starts of length, but I think don't think that is necessary for its simplistic purposes.
However, why not use WAR? Well, right, the more you pitch the more WAR you get. Okay, why not WAR/GS? Oh, because that would look terrible.
But what if you take advanced metrics in creating an adjusted quality start? Obviously that isn't something that you calculate while you sit at the game, thus the need for the simple QS, HQS, FS; however, aQS, aHQS, aNS, aLS, and aPS (adjusted Quality Start, High Quality Start, Neutral Start, Losing Start[????? there's a better term I'm sure], Poor Start), would be be calculated based upon things such as park factor in the park of the start (4R in 9 IP isn't a QS in Petco. In earlier years of Coors that might even be a HQS) to determine how often the average team wins in a start with X runs in Y IP. You would set a threshold for differentiating between the different starts (in terms of how often a your team would win the game--I dunno, maybe 45%-55% is aNS, above 70% is aHQS, between 55% and 70% is aQS, 33% - 45% is aLS and below 33% is aPS.
Anyway, then when comparing pitchers we would look at their percentages of each kind of adjusted Start. Kind of like game score but obviously without game score's obvious biases, and less misleading than the suggestion for the simple expansion of quality start suggested above (allowing for a more telling comparison of Ubaldo Jimenez' and Matt Latos' starts). The only other question would be whether to try to 'refine' it further by considering not how often the average team wins in a particular park when the starting pitcher gives up a set number of runs in a set number of innings, but how often that pitcher's team would win given their current run-scoring prowess (or lack thereof). But I think that's getting too cute. Because then the next person will say you should take into account the pitcher's opposing offense's ability to score runs, and that of his opposing pitcher's ability to suppress runs, and then of the respective bullpens, and so forth, and then it's just too convoluted and gets aware from the core idea of the quality start--a quick checklist for whether you gave your team a shot to win. The adjusted quality start would be a somewhat simplistic advanced calculation to make the same generic conclusions, but with a bit more relevancy.
A tiered system of types of starts based upon the average team's chances of winning the game being an interested statistic to look at for comparing pitchers; it could do things like spur debate on whether a dominant pitcher that also has a considerable number of poor stats is better than a pitcher that is more consistent but doesn't give his team as much of an edge when he pitches well. Also, since it's a per-start stat, it allows more insight, I think, than WAR does when two pitches have roughly the same WAR. If their WAR is the about the same and their percentages are about the same, okay, but if their WAR is about the same and their percentages have notable differences, whom to do you prefer?
Wow, this is long and rambling.
February 4th, 2011 at 4:37 am
"Since these games tend to be very low-scoring, I would think the likelihood of going to extra-innings would be somewhat higher than usual."
That's really faulty thinking. If the only runs in the game are unearned, logically most of the games are going to be shutouts. I think the first 9 I clicked on were in fact shutouts, but I don't have the time to waste clicking on all the others. Shame you couldn't have run this to show the game scores.
February 4th, 2011 at 4:42 am
Well there you go, I had to find out.
44 games, 43 shutouts.
Only non shutout. Game 39 CIN PHI 2002-04-14
That's the interesting talking point, or statistical aberration if you will. Not a small increase in extra inning games.
I really hope they aren't paying you for this.
February 4th, 2011 at 8:42 am
@40 Yes the majority of the games end in shutouts, but only because one team has to win eventually (no ties in baseball generally). Because we have narrowed it down to games with no earned runs, many of these game go to extras tied 0-0 and are won there on an unearned run or two.
February 4th, 2011 at 11:27 am
Random observations on a few of these no-ER games:
-- The Rockies played (and won) two of these games in a span of four days, 6/11/2008 (home to SF) and 6/14/2008 (@ CHW). The first game matched Ubaldo Jimenez and Tim Lincecum; both went 7 scoreless innings but got ND. The Rox won on an odd walk-off play: with 1 out and the bases loaded, Yorvit Torrealba flied to LF; Fred Lewis's throw to the plate was mishandled for an error by Bengie Molina, as Garrett Atkins scored the game-winner. Had Molina fielded the throw, Atkins would have been out, sending the scoreless game into extra innings -- which quite possibly would have been an historic first in Coors Field. (The Play Index does not let me determine that with certainty; however, it is a fact that there had never before been an extra-inning game in Coors Field that ended in a shutout. The first such game occurred 3 months later, on 9/14/2008, won by Colorado 1-0 in the 10th; Greg Maddux had gone 7 scoreless IP for LA on 2 hits and no walks in the 3rd-to-last game of his career.)
-- On 6/28/2008, the Angels lost 1-0 at the Dodgers, despite allowing no ER and no hits; the loss went to Jered Weaver, who went 6 IP and left for a PH in the 7th. The run was of the "extremely unearned" variety: Matt Kemp reached 1st on Weaver's own error, then stole 2nd and advanced to 3rd on a throwing error by C Jeff Mathis, and scored on a sac fly. In the past 20 seasons, Weaver is the only starting pitcher to take a loss when allowing no hits and no ER. And it was the only team no-hitter/loss since 1992.
-- The longest no-ER game in the past decade occurred on 5/31/2003, when the Cubs beat the Astros, 1-0, on a walk-off single by Sammy Sosa in the bottom of the 16th. The run was set up by an error by RF Brian Hunter, who had entered in the 9th as a defensive replacement. An interesting sequence took place in the 11th involving Sosa, who was having his last big year (40 HRs): With 2 out and a man on 2nd, Brad Lidge intentionally walked LHB Corey Patterson (who has having his only good season) to pitch to Sosa. It was the first time those two had ever faced off, and Lidge won this round by fanning Sosa, fitting Sammy with the "platinum sombrero" (?) -- 5 strikeouts in 5 times up. Sosa grounded out against Billy Wagner in the 14th, making him 0 for 6 before earning redemption in the 16th. (A year later, Sosa earned revenge on Lidge with a walk-off 10th-inning HR on 7/1/2004.)
February 4th, 2011 at 11:37 am
@Albert7: Why would you be insulting to Andy here? You've stamped YOUR opinion that one out of 44 games not being a shutout as the aberration, but there's really no reason that this should be more important than what Andy pointed out. I am, in fact, more interested in Andy's point - not to say that it's "objectively more interesting", but it's at least as valid, and then you go insult him. And so my question is: Why?
February 4th, 2011 at 12:15 pm
This is somewhat off the topic but didn't Ken Johnson of the Colt 45s lose a no hitter 1-0 on an unearned run. In fact I think maybe it was his own error.
February 4th, 2011 at 1:00 pm
didn't Ken Johnson of the Colt 45s lose a no hitter 1-0 on an unearned run. In fact I think maybe it was his own error.
Yes, Johnson did lose a no-hitter on an unearned run, on April 23, 1964. Johnson's own error allowed Pete Rose to reach second with one out in the top of the ninth, and Nellie Fox's two-out error allowed Rose to score the only run of the game.
February 4th, 2011 at 8:09 pm
John Autin #35-36
About WPA, I agree that in those individual circumstances you name, you have identified a limitation of the measure in that it attributes all of the probability added to one player. So I would agree that, for individual events or games, it's not a very good valuation of player worth.
However, in the long run, I would think that it is. Probability and random chance have a way of balancing out in the long run, and if Baker is truly a worse player than Abel, then Baker will strike out at bad times more over a season or career, and Abel will get good hits at good times more over a season or career. Or maybe that won't happen, which identifies another interesting trend, which is that maybe Baker is a better 'clutch' player. Certainly if he had a career full of reaching on dropped third strikes in the 9th inning, that wouldn't be a 'clutch' player so much as a lucky one, but equally certainly, the odds against that happening over and over again would be so astronomical that it simply wouldn't happen.
But, if there's a player that has a huge WPA and underwhelming other stats, then WPA can really highlight the fact that there's a statistical likelihood that that player is better 'in the clutch', which would be fascinating to identify. Or vice versa, it can identify a player who puts up phenomenal numbers but does nothing special when the game is on the line. It's sort of a check against the sports narrative cliche of a 'clutch' guy.
Anyway, like I said I agree with your examples, but in the long run I think it's a much different story, and WPA is a good measure of player 'value'.
February 4th, 2011 at 9:15 pm
Bells @46 -- You're probably right that WPA "luck" would tend to even out over time, at least to some degree. And granted, my Abel-Baker example was an extreme case.
But we're looking for a strong alternative to Wins & Losses. Why settle for something that has two of the same fundamental flaws as those stats? -- (a) that you need a fairly large sample before the luck starts to even out, and (b) that it tends to favor players on better teams? (I haven't fleshed out the latter point yet, but it does seem to me that WPA skews towards teams that are inherently more likely to win. I'll try to post some examples of this later.)
February 5th, 2011 at 12:29 pm
Hmm, yeah you're right in that context, I was just responding to it as a criticism of WPA in itself, not as a replacement for wins and losses.
Having a large sample is certainly a necessity, there's no disputing that. But I'd be careful about the 'skewing towards teams more likely to win' conclusion (which it seems you are being, to be clear). That's interesting and I'd like to see the data, but looking at the measure in and of itself, there's a question of causality that's wrapped up in it - after all, players with a high WPA would win more, all other things equal, and at some point those wins are caused by somebody. So yeah, it might 'skew', but that might be because good players create wins. I mean, at it's simplest WPA is a crude, but more nuanced breakdown of the 'Win-Loss' stat, to reflect more than the pitcher's role, right? It takes the smaller wins and losses (or rather, moments that contribute to the actual wins and losses) throughout the game and compiles them.
Back to the original question, then, I would say that yeah, WPA is a better measure than W-L for pitchers. Your 'quality start' analysis is also interesting. At the end of the day, I guess I'm a fan of a contextualization of a number of stats that highlight different angles than a gravitation towards one single stat, and W-L is obviously a highly overrated stat.
I'd be interested in looking into WPA a bit more and seeing how it might measure things better in the short run.
February 7th, 2011 at 7:41 am
[...] very own John Autin commented on Game Scores on a recent post: Consider two pitchers: -- Alfa goes 7 innings, allowing 0 runs on 6 hits, with 0 [...]